This new Project 3 in the UPCI Lung Cancer SPORE builds on both clinical and population based resource development and substantive research findings from the prior SPORE funding periods. Project 3 will pursue three translational aims to refine and validate a lung cancer risk prediction model and a diagnostic serum biomarker panel. In Specific Aim 1, Project 3 will construct a genetic risk index from single nucleotide polymorphism (SNP) genotype information, measure the index's association with lung cancer (Aim 1 A), and evaluate the index's contribution to our published lung cancer prediction model that incorporates demographic, cigarette smoking exposure, and clinical variables (Aim 1B). In Specific Aim 2, Project 3 will evaluate the lung cancer discrimination performance of a promising lung cancer diagnostic rule model constructed from the serum concentrations of 10 biomarker proteins (Aim 2A) and verify its performance in subjects with intermediate suspicion pulmonary nodules on screening computerized tomography (CT) (Aim 2B) and diagnostic performance independent of demographic, smoking, clinical, and genetic variables (Aim 2C). These aims use shared lung cancer cases (n=180) emerging from the Pittsburgh Lung Screening Study (PLuSS), a comparable lung cancer case series (n=548) derived from clinical sources, and controls [n=993 (Aim 2A) or n=482 (Aim 2B)] sampled from the PLuSS cohort. Collaboration with the Biostatistics and Bioinformatics Core will develop improved models based on these expansive datasets. A concluding translational aim uses independent comparison groups from the ACRIN/NLST and ACOSOG Z4031 cohorts to validate these prediction and diagnostic strategies (Aim 3). Completion of Aim 1 will produce a lung cancer risk prediction model that capably selects individuals for primary chemoprevention or enhanced screening. Aim 2 will yield a lung cancer diagnostic test to improve early lung cancer detection in the contexts of clinical trials of primary chemoprevention and management of high-risk subjects with CT-detected nodules. Lastly, Aim 3 will evaluate the performance of both the lung cancer prediction and diagnostic models in two large and similar case-control cohorts for independent validation and refinement of the final models.